Determination of Parabens in Pharmaceutical Formulations by Solid

The recently discovered estrogenic effects of certain synthetic chemicals, including the parabens, and their subsequent effects on the endocrine syste...
2 downloads 5 Views 415KB Size
Anal. Chem. 2005, 77, 5938-5946

Determination of Parabens in Pharmaceutical Formulations by Solid-Phase Microextraction-Ion Mobility Spectrometry John K. Lokhnauth and Nicholas H. Snow*

Department of Chemistry and Biochemistry, Seton Hall University, 400 South Orange Avenue, South Orange, New Jersey 07079

Solid-phase microextraction (SPME) coupled with ion mobility spectrometry (IMS) was used for the detection and quantitation of 4-hydroxybenzoate preservatives, methylparaben, ethylparaben, propylparaben, and butylparaben, in commercial pharmaceutical products. For the first time, SPME-IMS is described for the simultaneous detection, separation, and quantitation of multiple analytes in complex matrixes. The parabens are extracted from the samples using SPME, and the analytes on the fiber are heated by the IMS desorber unit and vaporized into the drift tube. The four preservatives differing only by a methyl group were separated in less than 18 ms. The analytical procedure was optimized for fiber coating selection, extraction time, sample pH, sample volume, ionic strength, and IMS conditions. Separation characteristics such as resolution, theoretical plates, and drift times of the parabens were also evaluated based on the direct interfacing of SPME to IMS. The conditions were tested using six over-the-counter topical products containing various combinations of preservatives. Analysis of the samples by SPME-IMS using benzyl paraben as an internal standard yields good comparison to an HPLC method, thereby reinforcing the applicability of this technique as a method for routine analysis. Limits of detection were 10 ng/mL for methylparaben and ethylparaben and 5 ng/mL for propylparaben and butylparaben. Good linearity range and reproducibility of less than 8% were obtained. Parabens are commonly used preservatives in topical pharmaceutical preparations, cosmetics, skin care products, medications, and foods and industrially in oils, fats, shoe polishes, textiles, and glues. Two or more paraben esters are often found in one product so it is useful to test sensitivity and selectivity with a paraben mix, such as methyl-, ethyl-, propyl-, and butylparaben as there is a high incidence of cross-reaction between the esters. Leave-on products such as facial makeup and skin lotions are of greatest concern because of the long exposure time and opportunity for migration via the skin into the bloodstream. The recently discovered estrogenic effects of certain synthetic chemicals, including the parabens, and their subsequent effects on the * To whom correspondence should be addressed. E-mail: snownich@ shu.edu.

5938 Analytical Chemistry, Vol. 77, No. 18, September 15, 2005

endocrine system of humans and wildlife, is of growing concern, especially in relation to the risk of breast cancer.1,2 For many years, parabens were considered among those preservatives with low systemic toxicity, primarily causing allergic reactions. However, as people have become aware that some synthetic chemicals mimic the female hormone estrogen, the understanding of the toxic effects of both synthetic and natural substances has changed. Significant effort has been invested to develop analytical techniques for active ingredients and preservatives in pharmaceutical products. These methods often require extraordinary care in sample collection, preparation, and analysis since only small amounts of substances might be available in complex matrixes that are difficult to extract or isolate. The demand also exists for rapid, inexpensive, and convenient analytical methods in quality control and assurance during the manufacture of drugs. More efficient methods of analysis could expedite the process of performing production line monitoring and final testing. The popularity of the parabens as preservatives in pharmaceutical formulations has led to a regulatory requirement for their rapid detection and quantitative analysis in development, manufacturing, and quality control laboratories. Parabens are traditionally analyzed by gas chromatography(GC) or high-performance liquid chromatography (HPLC).3-7 Derivatization as silyl- or fluoroacetyl derivatives combined with extraction and sample cleanup may be necessary for separation and quantitation in GC analysis. HPLC techniques, most commonly used for these compounds, often require extraction steps and long run times involving gradient elution to remove the many nonpolar compounds typically used in topical formulations. There has also been limited use of electrophoretic methods for these assays.8-11 (1) Routledge, E. J.; Parker, J.; Odum, J.; Ashby, J.; Sumpter, J. P. Toxicol. Appl. Pharmacol. 1998, 153, 12-19. (2) Harvey, P. W. J. Appl. Toxicol. 2003, 23, 285-288. (3) Cantwell, F. F. Anal. Chem. 1976, 48, 1854. (4) Thomassin, M.; Cavalli, E.; Guillaume, Y.; Guinchard, C. J. Pharm. Biomed. Anal. 1997, 15, 831-838. (5) Wang, S.; Chang, C. Anal. Chim. Acta 1998, 377, 85-93. (6) Akhtar, M. J.; Khan, S.; Roy, I. M.; Jafri, I. A. J. Pharm. Biomed. Anal. 1996, 14, 1609-1613. (7) Sottofattori, E.; Anzaldi, M.; Balbi, A.; Tonello, G. J. Pharm. Biomed. Anal. 1998, 18, 213-217. (8) Mahuxier, P. E.; Altria, K. D.; Clark, B. J. J. Chromatogr., A 2001, 924, 465-470. (9) Baalbaki, B.; Blanchin, M.; Fabre, H. Anal. Chim. Acta 2002, 463, 15-20. (10) Labat, L.; Kummer, E.; Dallet, P.; Dubost, J. P. J. Pharm. Biomed. Anal. 2000, 23, 763-779. 10.1021/ac0504013 CCC: $30.25

© 2005 American Chemical Society Published on Web 08/17/2005

Solid-phase microextraction (SPME) is a simple and effective sorptive sample preparation technique that has been employed in numerous applications for forensic, environmental, food, and pharmaceutical samples.12-21 However, SPME has not been previously utilized for paraben analysis. However, Karaisz and Snow used SPME for the analysis of degradants of other preservatives.22 Both headspace and direct extraction have been explored, with the primary instrumental techniques being gas and liquid chromatography. Orzechowska et al. have previously used SPME in combination with ion mobility spectrometry (IMS) to detect heroin and cocaine by headspace analysis.23,24 However, SPME-IMS was only used for semiquantitative purposes and has not been optimized or applied for multicomponent analytes in complex samples. Electrospray ionization and matrix-assisted laser desorption/ionization have also been investigated as interfacing techniques for SPME with IMS/MS.25,26 We have previously reported the development and optimization of SPME-IMS for the quantitative screening of ephedrine in urine.27 In this study, we show that paraben analysis in pharmaceutical products can be done in minutes with minimal sample cleanup and no solvents using SPME for one-step sample extraction/concentration and IMS for rapid multicomponent analysis. This technique has broad potential applicability beyond the current study. EXPERIMENTAL SECTION Chemicals and Materials. Methylparaben (MP), ethylparaben (EP), propylparaben (PP), butylparaben (BP), and benzylparaben standards were obtained from Sigma (St. Louis, MO). Hydrochloric acid, sodium hydroxide, sodium chloride, and ACS reagent grade acetonitrile were purchased from J.T. Baker (Phillipsburg, NJ). Ultrapure water used in all experiments was obtained from a Milli-Q unit (Millipore, Bedford, MA) installed in the laboratory. The SPME devices were purchased from Supelco Inc. (Bellefonte, PA) and used with the following extraction fibers: 65-µm poly(dimethylsiloxane)/divinylbenzene (PDMS/DVB), 65-µm polyacrylate, 7-µm poly(dimethylsiloxane) (PDMS), 100-µm PDMS, and 50/30-µm DVB/Carboxen/PDMS (DVB/CAR/PDMS). The (11) Huang, H.; Lai, Y.; Chiu, C.; Yeh, J. J. Chromatogr., A 2003, 993, 153-164. (12) Arthur, C. L.; Pawliszyn, J. Anal. Chem. 1990, 62, 2145. (13) Okeyo, P.; Snow, N. J. Microcolumn Sep. 1998, 10 (7) 551-556. (14) Yu, X.; Yuan, H.; Gorecki, T.; Pawliszyn, J. Anal. Chem. 1999, 71, 29983002. (15) Dewsbury, P.; Thorton, S.; Lerner, D. J. Environ. Sci. Technol. 2003, 37, 1392-1397. (16) Mejias, R.; Martin, R.; Moreno, M.; Barroso, C. J. Chromatogr., A 2003, 995, 11-20. (17) Raikos, N.; Christopoulou, K.; Theodoridis, G.; Tsoukali, H.; Psaroulis, D. J. Chromatogr., B 2003, 789, 59-63. (18) Gentilli, S.; Torresi, A.; Marsili, R.; Chiarotti, M.; Macchia, T. J. Chromatogr., B 2002, 780, 183-192. (19) Yuan, H.; Pawliszyn, J. Anal. Chem. 2001, 73, 4410-4416. (20) Zuccarello, J.; Ganske, J.; Green, D. Chemosphere 2003, 51, 805-810. (21) Chen, J.; Pawliszyn, J. Anal. Chem. 1995, 67, 2530-2533. (22) Karaisz, K., Snow, N. J. Microcolumn Sep. 2001, 13, 1-7. (23) Orzechowska, G.; Poziomek, E.; Tersol, V. Anal. Lett. 1997, 30, 14371444. (24) Orzechowska, G.; Poziomek, E.; Tersol, V.; Homstead, J. Proc. SPIE 1997, 2937, 8-18. (25) McCooeye, M.; Mester, Z., Ells, B.; Barnett, D.; Purves, R.; Guevremont, R. Anal. Chem. 2002, 74, 3071-3075. (26) Tong, H.; Sze, N.; Thomson, B.; Nacson, S.; Pawliszyn, J. Analyst 2002, 127, 1207-1210. (27) Lokhnauth, J. K.; Snow, N. H. J. Sep. Sci. 2005, 28, 612-618.

extraction vials (5 mL) were obtained from VWR (South Plainfield, NJ). The over-the-counter topical products, including three creams, a lotion, a solution, and an ointment, were purchased from neighborhood pharmacies. For the HPLC analysis, a Symmetry C18 5-µm 3.9 × 250 cm HPLC column was purchased from Waters Corp. (Milford, MA). IMS Parameters. IMS was performed using an Ionscan LS (Smiths, Warren, NJ) programmed in the negative mode, using 4-nitrobenzonitrile as a calibrant. Purified air was used as the drift flow gas, and it was doped with the reactant hexachloroethane in the reaction region of the drift tube to suppress signals from interferences that have a weaker electron affinity. The desorption temperature was set at 270 °C, the inlet temperature at 270 °C, the drift tube temperature at 115 °C, and the air flow rate set at 400 mL/min. Spectra were collected after a 1-ms delay with a shutter grid width of 0.2 ms. The scan period was set to 30 ms, and desorption time was set at 30 s. Methods. Standard stock solutions of the four parabens were prepared at a concentration of 0.1 mg/mL by dissolving 10 mg of MP, EP, PP, and BP in 2 mL of methanol and diluting to 100 mL with purified water. The internal standard, benzylparaben, was prepared similarly at a concentration of 0.1 mg/mL. The working standard solution was prepared by adding 50 µL of stock standard and 25 µL of internal standard solution into 10-mL flask and brought up to volume with water, yielding a concentration of 0.5 µg/mL parabens and 0.25 µg/mL of internal standard. Aliquots of 3 mL of the standard solution were transferred to 4-mL vials fitted with PTFE-lined silicone septa, and 600 mg of sodium chloride was dissolved producing a 20% ionic strength solution. Samples were prepared by dissolving 50 mg of the topical products into 100 mL of water and vortexing for 1 min. Internal standard solution was added to yield a concentration of 0.25 µg/mL and the ionic strengths were adjusted to 20%. SPME was performed on these solutions using a 50/30-µm DVB/CAR/PDMS fiber, which was conditioned according to the manufacturer’s instructions prior to analysis. The SPME fiber was exposed directly into the samples, and the extraction conditions were optimized for coating selection, exposure time, pH, and ionic strength, as described in the text. After sampling, the fiber was withdrawn into the needle and the SPME device was transferred to the IMS for thermal desorption and analysis. The extracted analytes were desorbed into the IMS drift tube by depressing the plunger on the SPME holder to expose the fiber. The exposed fiber was placed on the desorption tray in the center of the sampling region, and the tray assembly was slid all the way to the injection position, where the desorber rises, sealing the SPME fiber against the heated IMS inlet. Air was drawn through the sampling region at 400 mL/min to transfer the analytes from the fiber into the IMS drift tube for detection. This is depicted schematically in Figure 1. The topical products labeled as cream A, cream B, cream C, solution A, lotion A, and ointment A, containing various combinations of the preservatives, were prepared in water and quantitated using the internal standard method. HPLC analyses were also performed on the creams, lotion, solution, and ointment by dissolving 1 g of sample into 10 mL of methanol. A standard extraction procedure commonly used in the pharmaceutical industry was followed, which includes vortexing at high speed Analytical Chemistry, Vol. 77, No. 18, September 15, 2005

5939

Figure 1. (A) Schematic diagram of the SPME-IMS interface, showing the location of the SPME device. (B) Dynamics of the desorption process: analytes in the fiber coating. (C) Dynamics of the desorption process: analytes evaporating from the coating and entering the flow stream. Note the fiber is oriented 90° differently to the flow than in traditional SPME-GC.

for 5 min, heating in a 90 °C water bath for 5 min, followed by cooling in an ice bath for 5 min. The standard solutions were prepared by dissolving 10 mg of the parabens in 50 mL of methanol yielding a concentration of 0.2 mg/mL. The samples and standard preparations were filtered through a 0.2-µm filter and subsequently analyzed on an Agilent 1100 HPLC (Wilmington, DE). The analyses were performed by injecting 25-µL samples and standards, using a flow rate of 1.5 mL/min of 50:50 acetonitrile/water mobile phase and a silica-based C18 column with a 5-µm particle size and dimensions of 250 mm × 4.6 mm. Detection was done using ultraviolet spectrometry at 254 nm. RESULTS AND DISCUSSION Ion mobilities are determined from ion velocities that are measured in the drift tube of the spectrometer. The sample is heated by the desorber unit and vaporized into the instrument inlet. The vaporized sample is conveyed to the reaction region of the spectrometer, by dry purified air, where the reactant and calibrant ions are added. A 63Ni β emitter initiates a charge-transfer cascade in the reaction region that produces both positive and negative ions. In the negative ion mode, the gating grid and countercurrent drift gas inhibits cations and neutral molecules from entering the drift region of the spectrometer.28 5940

Analytical Chemistry, Vol. 77, No. 18, September 15, 2005

For the production of negative ions, there are three possible reactions that can occur. These include ion transfer, charge transfer, or dissociative charge transfer. In the negative mode, hexachloroethane, is used as the reactant substance. Ionization selectivity is obtained for compounds whose electron affinities are greater than that of the reactant ion through an equilibrium shift that is determined by the relative electron affinities of the reactant and analyte. For identification purposes, reduced mobilities (K0) are often used instead of drift time to correct for instrumental variations. Reduced mobilities are calculated for negative ion mode using 4-nitrobenzonitrile ion as the calibrant. The reduced mobilities of unknown peaks during analysis can be calculated using eq 1, where K0 is the reduced mobility (in units of

K0unknown ) K0calibranttcalibrant/tunknown

(1)

cm2 V-1 s-1) and t is the drift time of the calibrant and the unknown. (28) Eiceman, G. A.; Karpas, Z. Ion Mobility Spectrometry, CRC Press: Boca Raton, FL, 1994.

Figure 2. Extraction profile of parabens in cream B, sampled directly by five SPME fibers.

Operating at ambient pressure offers many advantages for IMS. These advantages include simple, robust, and miniaturized instrumentation that can be used for on site analysis. IMS is also a sensitive technique with real-time monitoring capabilities, short analysis times, and low detection limits. The consequence of electron transfer is that molecules form ions that maintain much of their original shape and size. Because the energetics of the APCI processes is weak, ions seldom dissociate or fragment in the reaction region of the spectrometer. The advantages of SPME with respect to selectivity and sensitivity render it an ideal sample preparation technique to combine with IMS. Different SPME fibers have different selectivities and sensitivities toward the target analyte. Therefore, different fibers must first be evaluated and then the optimal one selected for further studies. The choice of an appropriate coating is essential for the SPME method. Five types of fibers were studied: 65-µm PDMS/ DVB, 65-µm polyacrylate, 7-µm PDMS, 100-µm PDMS, and 50/30-µm DVB/CAR/PDMS. A 100-mg sample of cream B, containing methylparaben and propylparaben was dissolved in 100 mL of water. Ethylparaben and butylparaben were spiked into the cream solution to yield a concentration of 1 µg/mL of these two substances. The sample concentration and extraction parameters were kept the same for each study conducted on the five fibers. Figure 2 displays the extraction profiles of the parabens in a cream matrix by different fiber coatings extracted for 15 min. This figure shows that the 50/30-µm DVB/CAR/PDMS fiber exhibit the highest sensitivity to parabens for the 15-min duration. The difference in extraction performance among the fibers reflects variations in the polarity of the individual fibers and relative sorption affinity for the relatively polar hydroxybenzoates.The 7-µm PDMS fiber exhibited poor sensitivity toward methyl-, ethyl-, and propylparaben due to its nonpolar and low-capacity characteristics. As expected, butylparaben, which is relatively the most nonpolar compound in the group, exhibited better responses with the nonpolar 100-µm PDMS phases. The polyacrylate fiber produced poor responses for all analytes. The PDMS/DVB fiber, recommended for polar volatiles, exhibited a higher sensitivity for the analytes of interest in comparison with the polyacrylate fiber. Compared to the nonpolar liquid PDMS and the polyacrylate

coatings, methyl-, ethyl-, and propylparaben sensitivities could be increased considerably when the DVB/CAR/PDMS fiber was used. The DVB/CAR/PDMS fiber is generally recommended for extraction of relatively polar analytes and hence exhibited good sensitivity for parabens in comparison with the other fibers. The relative polarities of the four parabens are also reflected by the dramatic increase of methylparaben and ethylparaben responses with the more nonpolar DVB/CAR/PDMS fiber. For all further studies, the 50/30-µm DVB/CAR/PDMS fiber was chosen and the remaining fibers were not investigated further. Since SPME is an equilibrium extraction technique, the amount of parabens extracted by the fiber under a given set of conditions is dependent on the equilibration time, which is reached when a further increase of the extraction time does not result in a significant increase in detector response. The efficiency of paraben extraction by SPME was investigated using an extraction recoverytime curve. The study was done by comparing the response of known concentrations of parabens as a function of contact time with the DVB/CAR/PDMS fiber. Six replicate samples, prepared as described above, were extracted at 5, 10, 15, 30, 45, and 60 min. As seen in Figure 3, the amount of parabens adsorbed to the fiber leveled off after ∼15-min extraction time. An optimum sample extraction time of 15 min was therefore chosen to achieve maximum sensitivity without unduly extending the analysis time. The effect of pH on the SPME extraction of parabens was studied by adjusting the pH of the samples with 0.1 N HCl and 0.5 N NaOH. The pH of the cream B sample is ∼6, and the adjustments were made to pH 4, 5, 7, 8, and 9, followed by extraction with the DVB/CAR/PDMS fiber. As shown in Figure 4, little change was seen in the response when the pH was varied from 4 to 8. At pH higher than 8, the responses decreased noticeably. Therefore, pH adjustments were not made for all further analyses. Ionic strength can also be an important factor in SPME. Studies were done by comparing the responses of the parabens as a function of sodium chloride concentration. The responses of parabens in the cream matrix were investigated by performing extraction studies on samples where the ionic strengths of the solutions were adjusted with sodium chloride. Extractions were performed on 3-mL aliquots of spiked cream B sample with 0, 1, Analytical Chemistry, Vol. 77, No. 18, September 15, 2005

5941

Figure 3. MP, EP, PP, and BP responses as a function of sample extraction time by SPME.

Figure 4. MP, EP, PP, and BP responses as a function of sample pH by SPME.

Figure 5. MP, EP, PP, and BP responses as a function of sample ionic strength by SPME.

5, 10, 20, and 30% (w/w) sodium chloride added. The results are shown in Figure 5. Generally, an increase in the ionic strength of the sample will improve the sensitivity of some analytes by increasing the distribution coefficient, which leads to greater 5942

Analytical Chemistry, Vol. 77, No. 18, September 15, 2005

sensitivity. This phenomenon is usually more apparent with polar analytes; the peak area response of MP and EP increased significantly as the ionic strength is increased. There was also a small increase in response for PP as the ionic strength approached

Figure 6. Plasmagrams of (A) a standard solution containing 0.5 µg/mL MP, EP, PP, and BP and 0.25 µg/mL benzylparaben prepared in water. (B) A blank sample showing the large peaks produced from the reactant and also the lack of interferences in the analytes drift regions. (C) Cream C containing the parabens determined at concentrations of 2.05 mg/g MP, 0.39 mg/g EP, 0.21 mg/g PP, and 0.48 mg/g BP, along with the spiked internal standard. The peaks observed at approximately 10.5, 17.5, and 19 ms are due to sample composition substances present in cream C.

20%. The peak response of BP increased initially at 1% but decreased as the salt concentration increased further. An optimum sample ionic strength of 20% was chosen as a compromise to achieve adequate extraction efficiency for all of the compounds. In SPME, the amount of analyte extracted on the fiber (n) depends on the fiber volume (Vf), the distribution constant Kfs, and the sample initial concentration [A]o, and it may be dependent on the sample volume (Vs).

n ) KfsVfVs[A]o/(KfsVs + Vf)

(2)

Therefore, studies were performed to determine the optimum sample volume for paraben extraction. SPME were performed on 0.5 µg/mL standard, with volumes at 1, 3, 5, and 10 mL. No noticeable increase in the amount extracted was observed for the four analytes as the sample volume increased from 1 to 10 mL. Analytical Chemistry, Vol. 77, No. 18, September 15, 2005

5943

The constant responses over the volume range indicate that the product of distribution constant and sample volume is much greater than the fiber volume, so most terms divide out of eq 1, leaving the amount extracted dependent only on the fiber volume and the initial analyte concentration. For simplicity and ease of sample handling, a 3-mL extraction volume was chosen for all analyses. A plasmagram of a standard solution spiked with MP, EP, PP, BP, and internal standard, analyzed by SPME-IMS, is shown in Figure 6A. The plasmagram shows sharp Gaussian-shaped peaks for all analytes with separation for the five parabens achieved in less than 18 ms. The plasmagram is indicative of little or no fragmentation of the analyte ions, formation of ion clusters, or decomposition reactions in the drift tube. The Ionscan LS uses a reactant substance, hexachloroethane, to suppress responses from interferences that have a weaker electron affinity. In the negative mode, the hexachloroethane produces chloride ions, which augment the ionization process. The three peaks present from 7 to 9 ms, also present in the blank injections, shown in Figure 6B, are produced from the hexachloroethane reactant ions and their fragments. A plasmagram of cream C containing the four analytes along with the internal standard extracted by SPME is shown in Figure 6C. The peaks observed at approximately 10.5, 17.5, and 19 ms are due to additional ionizable substances present in cream C. The separation efficiency of the SPME-IMS method may be measured in terms of peak-to-peak resolution, Rpp, defined on the basis of separation of pairs of adjacent peaks, as in chromatography.29

Rpp )

2(td2 - td1) wb1 + wb2

(3)

The resolutions were calculated using eq 2, where td2 and td1 are the drift times of the two adjacent peaks, and wb1 and wb2 are their respective widths at the base. The resolution values calculated for the parabens were greater than or equal to 1, indicative of good separation for an IMS method. This separation power is interesting, considering that the four parabens of interest differ by just a methyl group. The separation efficiency of IMS, also referred to as the theoretical plates (N), is similar to that normally used in chromatography, and is a function of the drift tube voltage, temperature, number of charges on the ion, initial pulse width, length of the drift tube, and mobility of the ion of interest. The theoretical plates were calculated for the parabens, using eq 4,

N ) 5.55(td/wh)2

(4)

where wh is the peak width at half the maximum height. The results calculated for the five analytes determined to be greater than 10 000 theoretical plates, are summarized in Table 1. As IMS theory predicts, smaller ions have shorter drift times compared to larger ions, because they travel faster through the drift tube. The drift times needed by the ions to reach the detector are proportional to their masses but inversely proportional to their characteristic reduced ion mobilities. Many attempts have been (29) Asbury, G. R.; Hill, H. H. J. Microcolumn Sep. 2000, 12, 172-178.

5944 Analytical Chemistry, Vol. 77, No. 18, September 15, 2005

Table 1. IMS Drift Times, Reduced Mobilities, Resolutions, and Theoretical Plates of Paraben Analysis compound

drift time (ms)

reduced mobility (cm2 V-1 s-1)

resoltn

theor plates

methylparaben ethylparaben propylparaben butylparaben benzylparaben

12.746 13.538 14.295 15.053 16.140

1.4412 1.3569 1.285 1.2204 1.1382

n/a 1.09 1.04 1.00 1.31

11163 11778 13023 11342 12035

made to demonstrate that the masses of ions in a homologous series may be determined from ion drift time measurements. Most of the mass-mobility investigations were done on the homologous series of primary and tertiary aliphatic and aromatic amines. Good correlation has been obtained for mass-mobility measurements using a rigid sphere model involving parameters such as ion interaction potential, ion collision cross section, and various mathematical correction factors.30,31 A plot of Ko versus molecular weight for the four parabens of interest is shown in Figure 7 and exhibits good linear correlation with an r2 value of 0.996. This mass-mobilty correlation, observed without applying any of the traditional model calculations, indicates that the benzoate ions produced most likely do not form clusters and are probably molecular ions of the parabens. This correlation can be a very useful tool in estimating molecular weight of unknown ions based on their drift times in IMS. Quantitative linearity was investigated by determining calibration curves for the four parabens over the concentration range 0.05-1.0 µg/mL. The line of best fit for the relationship between the peak response and the concentration of each analyte in the standard solution was determined by linear regression, and the method was acceptably linear for all four analytes over the full range. The calibration curves are presented in Figure 8 and the correlation coefficients are presented in Table 2, together with the method precision, quantitation, and detection limits. The limits of detection and limits of quantitation (LOQ) were estimated from the signal-to-noise ratio of 3:1 and 10:1, respectively. Reproducibility of the method was evaluated and relative standard deviations (N)5) of less than 8% were found, as shown in Table 2. The lower end of the linear range, shown in Figure 8 differs from the LOQ estimated from the signal-to-noise ratio, due to some nonlinear behavior exhibited at lower concentrations. The relative standard deviations (N ) 5) for drift times were 0.5% for all analytes, indicating good repeatability, which is critical for analyte identification purposes in IMS. The parameters for linearity, range, reproducibility, limits of quantitation, and limits of detection reported here for a comprehensive method will vary for different products depending on the matrix compositions, interferences, ion competition, and suppression and, therefore, should be validated for each application before routine use. SPME-IMS was applied to various types of commercial topical pharmaceutical formulations. The samples designated as cream A, cream B, cream C, solution A, lotion A, and ointment A, containing various combinations of parabens, were tested using benzylparaben as an internal standard. Duplicate samples were (30) Berant, Z.; Karpas, Z. J. Am. Chem. Soc. 1989, 111, 3819-3824. (31) Karpas, Z. Anal. Chem. 1989, 61, 684-689.

Figure 7. Reduced mobilities vs molecular weight for the MP, EP, PP, and BP ions. The mass mobility correlation exhibits an r2 of 0.996 for the parabens.

Figure 8. Calibration curves for MP, EP, PP, and BP by SPME-IMS. Table 2. SPME-IMS Method Summary Showing Correlation Coefficient, Precision, Limits of Detection, and Limits of Quantitation for MP, EP, PP, and BP

methylparaben ethylparaben propylparaben butylparaben

correlation coefficient

precision (% RSD)

limit of detection (ng/mL)

limit of quantitation (ng/mL)

0.988 0.990 0.991 0.993

7.1 5.8 4.3 5.7

10 10 5 5

30 30 15 15

Table 3. Contents of MP, EP, PP, and BP Determined in Commercial Topical Formulations Determined by HPLC and SPME-IMS Shown with the Uncertainties of Each Determination quantitation (mg/g) sample cream a cream b cream c

tested, and the mean results (calculated in mg/g) are presented in Table 3. No agitation of the samples was done during extraction because agitation facilitated the extraction of additional compounds present in some samples, resulting in ion competition and paraben suppression. The equilibration time of 15 min without agitation was not excessively long and was considered practical for routine applications. To ensure that the fiber was clean and to avoid carryover, the SPME fiber was first exposed onto the IMS injector port as a blank run between injections. Since the fiber desorption time was set at 30 s, it was feasible to run blanks between analyses. HPLC is the most common technique that has been validated and used for routine analysis of these preservatives in pharma-

lotion a solution a ointment a

preservatives

by HPLC

by SPME/IMS

methylparaben propylparaben methylparaben propylparaben methylparaben ethylparaben propylparaben butylparaben methylparaben propylparaben methylparaben methylparaben butylparaben

0.98 ( 0.03 0.26 ( 0.01 1.98 ( 0.06 0.21 ( 0.01 1.92 ( 0.06 0.41 ( 0.01 0.22 ( 0.01 0.45 ( 0.01 1.42 ( 0.04 0.20 ( 0.01 1.65 ( 0.05 0.16 ( 0.01 1.48 ( 0.04

1.00 ( 0.07 0.24 ( 0.01 1.90 ( 0.13 0.19 ( 0.01 2.05 ( 0.14 0.39 ( 0.02 0.21 ( 0.01 0.48 ( 0.03 1.44 ( 0.10 0.18 ( 0.01 1.68 ( 0.12 0.18 ( 0.01 1.41 ( 0.08

ceutical topical formulations, so for evaluating accuracy of SPMEIMS, the six samples were tested by HPLC to substantiate the results obtained by SPME-IMS. The results for the HPLC method were calculated on a w/w basis and shown in Table 3, along with the SPME-IMS data. Analytical Chemistry, Vol. 77, No. 18, September 15, 2005

5945

In most cases, the difference obtained between the two methods for all of the products tested is not significant when the uncertainties for each method are taken into consideration. The standard deviations obtained by SPME-IMS appear larger than for HPLC. Because of varying proton/electron affinities and ion competition effects, when mixtures are injected directly into IMS, analytes can be completely missing or largely suppressed from plasmagrams until their concentrations become larger than the other compounds present in the sample. Historically, this limitation has been problematic when quantitative determination is necessary for complex samples. Therefore, it is noteworthy that significant discrepancies were not seen in the SPME-IMS results for parabens since these preservatives are a very small percentage of the total amount of substances present in the samples. As seen in Table 3, generally the larger differences between the two methods occur where the paraben concentrations are less than 0.5 mg/g. Additional optimization of the SPME extraction method for each matrix type can further minimize the discrepancies observed for some determinations. For samples containing higher levels of interfering compounds, additional extraction parameters can be adjusted to achieve more efficient paraben recoveries. We note that in a full method validation process for pharmaceutical application, a spike study using known concentrations of analyte added to known samples should be performed to further confirm quantitative accuracy. Quantitation by both SPME and IMS has been shown to be highly dependent on the matrix composition. Since ion suppression has traditionally been a major limitation in the application of IMS to detect and quantitate analytes in complex matrixes, the relative composition of the liquid matrix should be considered carefully when the data obtained using these methodologies are used. It was expected that, in the real samples tested, differences in proton affinities, vapor pressures, and relative composition would have resulted in preferential ionization of components and suppression of the paraben peaks in the mobility spectrum. However, the amounts of parabens recovered were not significantly affected due to the selectivity of the SPME extraction procedure and also the elimination of positive ions, obtained by running the IMS in the negative mode. Ion competition among the parabens is also minimized because of the similar functional groups and closeness in structures of the four analytes. The minimal matrix interferences and lack of ion suppression are reflected in the closeness of the results obtained by the chromatographic method. The great advantage of the SPME-IMS method is the ability to determine preservatives in the presence

5946

Analytical Chemistry, Vol. 77, No. 18, September 15, 2005

of other components in the sample with acceptable confidence without the need to perform long chromatographic separation, such that the time and cost of analysis per sample is significantly reduced. The closeness in quantitative results obtained from the two orthogonal methods strongly confirms the applicability of the proposed SPME-IMS method as a reliable technique that can be used for routine analysis of the parabens. CONCLUSIONS This paper describes the development of SPME-IMS as a new technique for the determination of methyl-, ethyl-, propyl-, and butylparaben in pharmaceutical cream, lotion, solution, and ointment samples. The method does not require elaborate sample preparation or chromatographic separation and provides a straightforward, fast, sensitive, and reliable method for detecting and quantifying the preservatives in complex pharmaceutical matrixes. The reproducibility of less than 8%, linear range from 0.05 to 1 µg/mL, and sensitivities of less than 10 ng/mL render SPMEMS very attractive for routine applications. Separation of the four analytes and internal standard with resolution of >1 and theoretical plates of >10 000 for all four parabens in less than 18 ms was obtained. The quantitation of preservatives in the real samples using HPLC and SPME-IMS indicates good agreement between the methods, with differences in paraben recoveries generally within experimental errors. Discrepancies between the two methods are observed in samples that contain high levels of interfering compounds with respect to parabens, which result in ion interferences. The main advantages of the SPME-IMS method include the reduction of labor-intensive extraction steps, elimination of solvent consumption in extraction and mobile-phase preparation, and shortened analysis time from hours to minutes. With appropriate validation for each matrix type, SPME-IMS can be easily applied to quality control, on site manufacturing in-process testing, and field analysis. ACKNOWLEDGMENT The authors thank Dr. Warren Wezel and Dr. James Romito for their support. Irwin Greene, Dr. Andy Anantha, and Dr. Norman Canfield are also gratefully acknowledged for their insights and assistance with ion mobility spectrometry. Received for review March 7, 2005. Accepted July 19, 2005. AC0504013